Skip to main content
Radar

Keep up with the models you depend on. Follow price changes, retirements, and API updates.Follow the models you depend on.

Follow model changes

ExploitGym

A controlled benchmark for evaluating whether AI agents can extend vulnerability-triggering inputs into working exploits.

Data verified 37 confirmed releases in the last 30 daysSee provider release alerts

Benchmark score on ExploitGym — September 10, 2026

We mirror the published score view for ExploitGym. GPT-6 Astra leads the public snapshot at 42.4%, followed by GPT-5.6 Sol (33.7%) and GPT-5.6 Terra (23.2%). We do not use these results to rank models overall.

10 modelsAgenticCurrentDisplay onlyUpdated September 10, 2026

Benchmark score table (10 models)

Score
1
GPT-6 AstraOpenAI · Closed
42.4%
2
GPT-5.6 SolOpenAI · Closed
33.7%
3
GPT-5.6 TerraOpenAI · Closed
23.2%
4
Claude Mythos PreviewAnthropic · Closed
17.5%
5
DeepSeek V4.1 FlashDeepSeek · Open weight
15.3%
6
GLM-5.3Z.AI · Open weight
15.0%
7
GPT-5.5OpenAI · Closed
13.4%
8
GPT-5.6 LunaOpenAI · Closed
12.4%
9
GPT-5.4OpenAI · Closed
6.0%
10
Muse Spark 1.1Meta · Closed
0.8%

The published ExploitGym snapshot places GPT-6 Astra first at 42.4%. The third row is 19.2 points behind. The broader top-10 range is 41.6 points, so the table still separates the published systems.

10 models have been evaluated on ExploitGym. The benchmark falls in the Agentic category. This category carries a 22% weight in BenchLM.ai's overall scoring system. ExploitGym is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.

About ExploitGym

Year

2026

Tasks

898 exploitation tasks

Format

Working exploit generation

Difficulty

Advanced cybersecurity exploitation

ExploitGym contains 898 containerized tasks sourced from real-world vulnerabilities across userspace programs, V8, and the Linux kernel. BenchLM normalizes successful exploit counts to percentage of the 898-task suite and keeps the benchmark display-only because it measures dual-use offensive cybersecurity capability.

BenchLM freshness & provenance

Version

ExploitGym 2026

Refresh cadence

Quarterly

Staleness state

Current

Question availability

Public benchmark set

CurrentDisplay only

BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.

FAQ

What does ExploitGym measure?

A controlled benchmark for evaluating whether AI agents can extend vulnerability-triggering inputs into working exploits.

Which model scores highest on ExploitGym?

GPT-6 Astra by OpenAI currently leads with a score of 42.4% on ExploitGym.

How many models are evaluated on ExploitGym?

10 AI models have been evaluated on ExploitGym on BenchLM.

Last updated: September 10, 2026 · BenchLM version ExploitGym 2026

Know when it’s worth switching models

The model to choose, the cheaper alternative, and the release we would wait on.

Read a sample issue

Join 2,000+ readers.

One email each week. Unsubscribe anytime.